Following on last years work--UPATSU (Understanding Protein Architecture Through Simulated Unfolding)--we have been computationally studying protein folding. There are two approaches. One is to computationally simulate the refolding of partially unfolded structures. The other is to grow a polypeptide amino acid by amino acid in the same way as it occurs in ribosomal translation. Molecular dynamics is done for 100 picoseconds after the addition of each amino acid. These computations for several known crystallographic proteins will have taken several monts of array processor time. The goal is to mesh the results from the two approaches and to thus obtain the ability to build a model of any new protein.